Satisficing, Optimization, and Adaptive Systems
JASON BROWNLEE
Technical Report 070305A
Complex Intelligent Systems Laboratory, Centre for Information Technology Research,
Faculty of Information Communication Technology, Swinburne University of Technology
Melbourne, Australia
jbrownlee@ict.swin.edu.au
Abstract- An aspect of complex adaptive systems is an
observation that they typically operate far from equilibrium and
optimality. This provokes investigation into frameworks into
such decision-making processes. Satisficing describes a rational
decision making process in economics where deciding agents
accept solutions that achieve a minimum level of satisfaction.
This theory differs from previously traditional rational decision-
making in which the agent seeks to maximise or optimize utility
from the choices faced. This work investigates some of the impact
this altered perspective on the decision making process has had
in economics, game theory, control theory, and evolutionary
biology. We see it is the latter case that that may be most
interesting to the study of complex adaptive systems and
resultant models, such as those used in biologically inspired
computation.
Keywords- Satisficing, Adaptation, Optimization, Evolution,
Algorithms, Complex Adaptive Systems
I. INTRODUCTION
Consider Complex Adaptive Systems (CAS) [14] as
systems of collectives of adaptive agents whose local
interactions result in dynamical non-linear effects and
emergent behaviours. According to Holland [14,15], and
others [35,36,38], such systems operate far from
equilibrium and optimality. This is an intriguing notion
given the ubiquitous adaptationist model of evolutionary
biology (a popular example of a CAS) as an
optimization process. That is, organisms are optimized
by evolution for their environment.
This work softly explores this theme by providing
background on the decision theory of good enough


solutions known as satisficing and a number of
interesting and notable works interpreting this notion in
the context of game theory, control theory, ecology, and
search. Satisficing is a notion suitable to describe the
operating far from optimality description of CAS, and a


final discussion considers a few implications of
satisficing in designing future artificial evolutionary and
immune system algorithms.
II. SATISFICING
The Nobel Prize laureate Herbert A. Simon coined
the term satisficing to describe the selection of a good


enough solution, the selection of a decision that meets a


minimum threshold or aspiration level, the selection of
which occurs in the context of incomplete information or
limited computation [7,8]. This formalism came out of
his work on bounded rationality , the idea of agents


CIS Technical Report 070305A March 2007
acting approximately rational, rather economically
rational in terms of utility. Simon had a problem with the
idea of the traditional economic and rational agent,
 
 
because in reality, agents do not always optimize their
decision making process. He drew from the field of
psychology to describe a characteristic decision maker .


In so doing, he outlined a scenario of rational choice:
(1) There are set of alternatives open to choice
(2) The relationships that determine payoffs
(satisfaction or attainment) are a function of the
alternatives that are chosen
(3) There is preference orderings among payoffs
(ordinal)
In this generalised scenario, the decision maker must
apply a strategy. Traditional economic strategies of the
time include: the max-min rule, the probabilistic rule,
and the certainty rule. Simon s problem was that these


strategies required information or computation that no
human could possess, and hence were unrealistic.
Satisficing: The selection of an acceptable or satisfactory solution that




meet an agents minimum aspiration-level or threshold , a threshold, under
 
 
which solutions are deemed unacceptable . A satisficing solution may or


may not be an optimal economic solution.
Figure 1 - Canonical definition of the term satisficing


Satisficing differs from a traditional rational
decision-making process in economics in which the
agent assigns a utility to each choice, and then makes a
decision based on the probability that each choice would
occur maximising utility. In contrast, satisficing
solutions are deemed either satisfactory (above the

agents aspiration level) or unsatisfactory (below the
agents aspiration level). An agent may decide on the
first, or any satisficing choice, the outcome being that
the agent is always satisfied, although the solution may
or may not be economically optimal. Such a strategy is
simple, computationally tractable, and a rational
strategy, although always selecting a satisficing strategy
may not be rational.
Simons satisficing decision-making and bounded


rationality have had a profound effect in many fields,
not limited to economics, game theory, management
theory, psychology, artificial intelligence and
evolutionary biology. Some interesting and notable
works follow.
Some general works include; a recent collection of
page 1 of 5
essays on satisficing decision making [25], satisficing in
economics and firms [33], decision theory on deciding
when a selected capacity has been learned and overt
learning should cease [34], and a philosophical treatment
of satisficing verses optimization [24].
Much work has been done to extend satisficing and
comparative rationality in game theory by Stirling, et


al. not limited to satisficing in games such as the
prisoners dilemma and battle of the sexes [41]. There is
a recent book dedicated to the topic of satisficing
games [40] including computer models and algorithms.


Stirling has also done work on decision-making with
regard to appropriate skill selection, as opposed to
selecting the single best skill [23], and so called
satisficing equilibria [39]. There has also been work by


others on the evolution of cooperation with regard to
satisficing strategies in [42], and resultant cooperation
given satisficing strategies in mutual interest games [3].
Satisficing has also proven a useful inspiration for
decision making in control systems. In particular is the
development of the so called satisficing control theory


for nonlinear systems controllers [18-20]. The satisficing
threshold was paired with epistemic utility theory as a


multi-objective approach to accepting and rejecting
propositions.
Satisficing controllers have been used in robot soccer
simulations [13], and so called action selection in robot


theory [27]. Sen [32] provides a collection of essays on
satisficing models of control when dealing with
uncertainty. Finally, satisficing has been applied to
design-to-criteria [37] as a multiobjective approach to


meeting the needs of clients, and in [17] as a theory to
describe the problem of human designers latching onto
one design (even if the design is sub-optimal), not
considering other designs.
ADAPTATION AND OPTIMIZATION
Thus far, satisficing has been superficially
demonstrated to be an interesting little pocket of
decision theory, it isn t until it is considered in the


context of ecology and evolutionary biology that the
interesting link to adaptive systems becomes apparent.

Bracketing satisficing with Darwinian [evolution]
may appear contradictory, for evolutionists sometimes
talk about survival of the fittest. But, in fact, natural
selection only predicts that survivors will be fit enough,
that is, fitter than their loosing competitors; it postulates
satisficing, not optimization

From [11] (pg 192) quoting Simon and Newell.
Adaptation by Darwinian evolution by natural
selection and is often phrased survival of the fittest , as


the optimization of the best or fittest organisms to



their environment. A well-known problem to ecologists
and evolutionary biologists is that there is no clear
fitness for natural selection to maximise. As Simon and
Newell point out in the above quote, perhaps rather than
the conventional optimizing adaptationist discussion of
evolution, a satisficing perspective may be more
insightful. Evolution phrased in this way would perhaps
be survival of the comparatively fit not the most fit,


where evolutionary fitness is measured as the trade-off
CIS Technical Report 070305A March 2007
of many different local and relative considerations.
A popular quote from Dennett from his work on
intention theory [4]:
...poor

old Mother Nature makes do,
opportunistically and short-sightedly exploiting
whatever is in hand-until we add: she isn t perfect, but
she does the best she can. Satisficing itself can often be
shown to be the optimal strategy when costs of


searching are added as a constraint .

Richardson considers the question of optimization in
the context of evolution ecology [30], and the tradition
of considering evolution as maximising adaptiveness of
an agent. In his treatment, he shifts the focus from
evolution as an optimizing process, to that of the
decision making component natural selection .


He contends that optimization models are simpler
than satisficing models and easier to assess. He further
contends that satisficing may be considered
optimization, although a more complex optimization in
many dimensions. Richardson discusses two points in
which optimization and satisficing differ: 1. Whether the
constraints defining an optimal solution are dynamically
significant , and 2. Structure of the environment . Here



he is referring to the consideration of a sequential
decision making process of satisficing where unlike
optimization, evolution is working with what it has
based on the organisms history, that alternatives for the
organism are not presented in parallel. He does not
propose that satisficing presents a major shift in thinking
about evolutionary ecology from optimization, rather a
minor change in perspective.
Satisficing Evolution: Satisficing provides decision making mechanism in
which to consider evolution by natural selection not as a process of seeking
maximal adaptiveness of an agent to its environment, but rather sufficient
 

adaptiveness . Such adaptation may or may not be optimal in the

consideration of all the constraints of the agent.
Figure 2 - A phrasing of evolution as a satisficing rather than optimizing
Also in the field of ecology, satisficing has been
proposed as a model for predator search (foraging) and
parental behaviour [5]. Complaints of such models,
when compared to conventional optimizing foraging
models is that they are considered both difficult to test
and perhaps a subset of optimizing models. Although
testable satisficing foraging models have been proposed
[26]. So called information-gap theory 1 has been


incorporated into satisficing foraging models supporting
experimental observations better than conventional
models.
III. SATISFICING IN SEARCH
Satisficing has influenced considerations in
computational search and optimization. This section
summarises some of the influence satisficing has had in
this field of research.
1
Information-Gap Decision Theory: Optimizing robustness to failure or
opportunity to windfall rather than conventional decision theory that
optimizes utility. The gap refers to what is known, and what needs to be
known.
page 2 of 5
 Simon, among many other things in his prestigious
career, was a founding contribution to the field of
artificial intelligence and heuristics. One of his many
contributions in this field was his ideas of satisficing
search [9]. This involves problems to which the


solutions are all or nothing , where there is no ordering


between solutions, but there are many ways in which to
achieve a solution. Thus constraints are placed on the
order in which paths are selected to be searched. Such
problems can be phrased as and-or trees. Some related
and extensions include [6] and [31]. The approach also
been used in learning while doing decision making for


seeking approximately optimal solutions [28].
Ho [43,44] proposed that satisficing provides an
interpretation of optimization of difficult and complex
problems where a good enough solution may suffice,


promoting the use of heuristic and approximate
techniques. That is a solution that is in the top n% of
optimal solution. Another satisficing interpretation in the
context of search is to that of multiple objective
optimization, that is the concurrent optimization of
multiple orthogonal objectives, not limited to [10], [12],
and [22].
Satisficing in Search
Threshold: A threshold is imposed with regard to the satisfaction or
suitability of a solution. Once a solution meets such a threshold, the search
may be considered satisfied.
Tradeoffs: More than one consideration (such as fitness or utility) is
evaluated when seeking a solution.
Comparative: from satisficing evolution, solution utility may be considered
comparatively rather than absolutely. This ordinal ranking may further be
reduced to the Boolean status of satisfactory and unsatisfactory solutions.
Presence of Optimality: also from satisficing evolution, unlike optimization
that assumes the presence of a global or locally optimum solution, satisficing
makes no such assumptions about the existence of an optimum where one
may or may not exist.
Figure 3 - A summary of the major interpretations of satisficing in search
Juric proposes a so called anti-adaptationist genetic


algorithm [21] based on satisficing ideas of generating
solutions with minimum threshold rather than optimize.
He claims that the optimizing adaptive method to search
may impede finding solution. He claims satisficing
almost as an antonym to optimizing adaptation as a


closer match to biology. He lists four main problems he
has with the adaptationist approach, claiming that
adaptation is the problematic idea that natural selection
chooses a particular design for a particular reason (best
possible result given the circumstance).








Juric s problems with adaptationist GA s








• Assumes an optimum exists to which the genotype should be
guided
• Assumes continued application beyond a minimum acceptance
level seeking the best possible solution
• Assumes that schemata that are currently useful will remain useful,
and that those that are useless will remain useless in the future
schemata theorem
Assume that building blocks that serve a purpose now will
•
continue to serve the same purpose in the future
Figure 4 - The four main problems Juric has with the adaptationist GA
His anti-adaptation approach is claimed to allow for


the use of evolved attributes for other purposes. It is not
CIS Technical Report 070305A March 2007
limited to optimization that limits variability, viability,
plasticity, flexibility in its drive for an optimum. His
approach encourages a minimum standard, robustness.
Claiming to encourage the adaptability of the structure,
rather than adapting the structure. No implementation is
provided of his conceptions.
Channon also investigates the same ideas of what
could be called satisficing in genetic algorithms with the
goal of novelty [1] and open-ended, persistent
evolutionary systems [2]. He claims that genetic
algorithms perform artificial selection with explicit


fitness function, what he calls abiotic selection . Such


systems are unable to perform long-term incremental
evolution with continual emergence. He investigates
what he calls natural or biotic selection in which there


is no explicit specification of what is desired; rather
evolution is interpreted as theory of local change and
satisficing resulting in continual emergent phenomena.
One cannot help but consider such an emergent
interpretation of evolution as a strong match to complex
adaptive systems theory.
IV. DISCUSSION
Satisficing has had an interesting but perhaps still not
fully exploited impact on search and optimization
theory, and may be an important aspect of complex
adaptive systems, and in biologically inspired
computation. It is the latter which I will briefly discuss
in this section, in particular in the context of
evolutionary computation and Artificial Immune
Systems (AIS).
The need to consider evolutionary algorithms as
complex adaptive systems rather than only as optimizers
is known, but infrequently heeded [16]. Such algorithms
can be used for optimization tasks with what Channon
calls artificial selection (like the artificial selection in


the breeding of horses or dogs), although perhaps the
satisficing interpretation of evolution (and perhaps other
inspirations for CAS and computational models) invites
investigation into other aspects of such systems.
This work has highlighted some potentially fruitful
avenues for continued and future research. An important
one is that of open-ended systems resulting in emergent
evolution. The specification of looser objective functions
or perhaps no objective functions where solutions are
discriminated as satisfactory or unsatisfactory. Such
systems promote adaptability rather than perform
adaptation, potentially providing future robustness in
uncertain environments.
Does satisficing provide a way to reconcile
evolutionary algorithms as a model of artificial rather
than natural selection, to reconcile approximate (less-
than-optimal) optimization?
With regard to the immune system of vertebrates,
satisficing points out the obvious, that the system does
not provide a perfect defence. If the system were an
optimal adaptation, then there would be no disease.
Nesse [29] discusses the mal-adaptations that result from
evolution, highlighting the immune system and disease
as a case study. He questions as to why natural selection
left us vulnerable to disease the evolutionary benefit,

page 3 of 5
and why bodies were not designed better.
The typical argument from evolutionary biologists is
that natural selection is an undirected process whose
influence is weak because of the stochastic nature of
genetic drift and mutations. We should not expect bodies
to be anywhere near optimal. This argument, while
good, is insufficient, because while many aspects of say
the human are less than optimal, manu others are
considered quite close to optimal. He proposes a
framework of six reasons why we are vulnerable to
disease, grouped into three propositions.
Possible reasons why humans are vulnerable to disease
(1) Selection is slow (time lag), results in:
(a) A mismatch between design and environment
(b) Competition with a pathogen or other organism
(2) Selection cannot solve some problems regardless of how much time it has,
results in:
(a) Tradeoffs
(b) Constraints particular to living organisms (path dependence)
(3) We misunderstand what selection shapes, results in:
(a) Traits that increase RS at the cost of disease vulnerability
(b) Adverse defences are readily mistaken for diseases
Figure 5 - Nesse's six reasons why humans may be vulnerable to disease
Clearly, the immune system is satisficing rather than
optimizing, providing a good solution to detecting and
neutralising antigen, not necessarily optimal. These
points raise many questions. More importantly, this line
of inquiry provokes interesting lines of research in CAS
and AIS such as:
What other constraints have evolution considered in
its evolution, what has natural selection traded off? What
would a perfect immune system look like? Is there a
mismatch between the requirements and the
implementation of the vertebrate immune system, and if
so what is it? Finally, what if rather than use artificial
selection and simulate the superficial aspects of the
immune system (such as clonal selection), what if we let
such a defensive system evolve via artificial emergent
evolution? What would a satisficing interpretation of
clonal selection theory reveal, given the theories
selectionist basis was inspired by Darwin s evolutionary


theory?
ACKNOWLEDGMENTS
Tim Hendtlass for his patience and for providing
useful feedback on drafts of this paper
REFERENCES
[1] A. D. Channon and R. I. Damper, "Evolving novel behaviors via
natural selection," Proceedings of Artificial Life VI, Los Angeles, USA, pp.
384-388, 1998.
[2] Alastair Channon, Evolutionary emergence: the struggle for
existence in artificial biota 2001. University of Southampton.
[3] Amit Pazgal, Satisficing leads to cooperation in mutual interests
games International Journal of Game Theory, vol. 26, pp. 439-453, Dec,
1997.
[4] D. Dennett, Intentional Systems in Cognitive Ethology: The
Panglossian Paradigm Defended The Behaviorial and Brain Sciences, vol. 6,
pp. 343-390, 1983.
[5] D. Ward, The role of satisficing in foraging theory OIKOS , vol.
63, pp. 312-317, 1992.
[6] Dan Geiger and Jeffrey A. Barnett, Optimal Satisficing Tree
Searches National Conference on Artificial Intelligence, vol. pp. 441-445,
CIS Technical Report 070305A March 2007
1991.
[7] Herbert A. Simon, A Behavioral Model of Rational Choice
Quarterly Journal of Economics, vol. 69, pp. 99-118, 1955.
[8] Herbert A. Simon, Rational choice and the structure of the
environment Psychological Review , vol. 63, pp. 129-138, 1956.
[9] Herbert A. Simon and Joseph B. Kadane , Optimal problem-
solving search: all-or-none solutions Artificial Intelligence , vol. 6, pp. 235-
247, Autumn, 1975.
[10] Hiroyuki Tamura, Tomohiro Shibata, and Itsuo Hatono,
"Multiobjective combinatorial optimization for performance evaluation by a
metaheuristic satisficing tradeoff method," International conference on
Advances in production management systems, Berlin, Germany, pp. 490-497,
2000.
[11] Hunter Crowther-Heyck. Herbert A. Simon: The Bounds Of
Reason In Modern America, John Hopkins University Press, 2005.
[12] I. Tamura, T. Shibata, S. Tomiyama, and I. Hatono, "A meta-
heuristic satisfying tradeoff method for solving multiobjective combinatorial
optimization problems-with application to flowshop scheduling," Proceedings
of the IEEE International Conference on Systems, Man, and Cybernetics
(IEEE SMC '99 Conference), Tokyo, Japan, pp. 539-544, 1999.
[13] Jay Packard, Satisficing applied to simulated soccer 2003.
Electrical and Computer Engineering, Brigham Young University.
[14] John H. Holland. Hidden Order: How Adaptation Builds
Complexity, USA: Addison Wesley Publishing Company, 1995.
[15] John Henry Holland. Adaptation in Natural and Artificial
Systems: An Introductory Analysis with Applications to Biology, Control, and
Artificial Intelligence, USA: MIT Press, 1975.
[16] Kenneth A. De Jong, "Genetic Algorithms are NOT Function
Optimizers," Proceedings of the Second Workshop on Foundations of Genetic
Algorithms (FOGA), Vail, Colorado, USA, pp. 5-17, 1992.
[17] L. J. Ball, L. Maskill, and T. C. Ormerod, Satisficing in
engineering design: causes, consequences and implications for design support
Automation in Construction, vol. 7, pp. 213-227, Jan, 1998.
[18] M. A. Goodrich, W. C. Stirling, and R. I. Frost, "A satisficing
fuzzy logic controller," Proceedings of the Fifth IEEE International
Conference on Fuzzy Systems, New Orleans, LA, USA, pp. 272-276, 1996.
[19] M. A. Goodrich, W. C. Stirling, and R. L. Frost, "A satisficing
approach to intelligent control of nonlinear systems," Proceedings of the 1996
IEEE International Symposium on Intelligent Control, Dearborn, MI, USA,
pp. 248-252, 1996.
[20] M. A. Goodrich, W. C. Stirling, and R. L. Frost, A theory of
satisficing decisions and control IEEE Transactions on Systems, Man and
Cybernetics, Part A, vol. 28, pp. 1083-4427, Nov, 1998.
[21] M. Juric, "An anti-adaptationist approach to genetic algorithms,"
Proceedings of the First IEEE Conference on Evolutionary Computation,
Orlando, FL, USA, pp. 619-623, 1994.
[22] M. Sakawa, K. Kato, and T. Shibano, "An interactive fuzzy
satisficing method for multiobjective multidimensional 0-1 knapsack
problems through genetic algorithms," Proceedings of IEEE International
Conference on Evolutionary Computation, Nagoya, Japan, pp. 243-246,
1996.
[23] Michael A. Goodrich, Wynn C. Stirling, and Erwin R. Boer,
Satisficing Revisited Minds and Machines, vol. 10, pp. 79-109, Feb, 2000.
[24] Michael Byron, Satisficing and Optimality Ethics, vol. 109, pp.
67-93, Oct, 1998.
[25] Michael Byron. Satisficing and Maximizing: Moral Theorists on
Practical Reason, USA: Cambridge University Press, 2004.
[26] P. Nonacs and L. M. Dill, Is satisficing an alternative to optimal
foraging theory Oikos, vol. 24, pp. 371 3751993.
[27] Paolo Pirjanian, Satisficing Action Selection Sensor Fusion and
Decentralized Control in Robotic Systems (SPIE Conference), vol. 3523, pp.
157-168, 1998.
[28] R. Greiner and P. Orponen, Probably Approximately Optimal
Satisficing Strategies Artificial Intelligence, vol. 82, pp. 21-44, Apr, 1996.
[29] Randolph M. Nesse, Maladaptation and Natural Selection The
Quarterly Review of Biology, vol. 80, pp. 62 702005.
[30] Robert C. Richardson , Optimization in Evolutionary Ecology
Philosophy of Science Association (PSA) , vol. 1, pp. 13-21, 1994.
[31] Russell Greiner, Ryan Hayward, Magdalena Jankowska, and
Michael Molloy, Finding optimal satisficing strategies for and-or trees
Artificial Intelligence, vol. 170, pp. 19-58, Jan, 2006.
page 4 of 5
 [32] Sandip Sen, "Satisficing Models," Published by The AAAI Press,
Menlo Park, California, USA, Technical Report SS-98-05, 1998.
[33] Sidney G. Winter, Satisficing, Selection, and The Innovating
Remnant Quarterly Journal of Economics, vol. pp. 237-261, 1971.
[34] Sidney G. Winter, The Satisficing Principle in Capability Learning
Strategic Management Journal, vol. 21, pp. 981-996, 2000.
[35] Simon A. Levin, Ecosystems and the Biosphere as Complex
Adaptive Systems Ecosystems, Biomedical and Life Sciences and Earth and
Environmental Science, vol. 1, pp. 431-436, Sep, 1998.
[36] Simon A. Levin, Complex adaptive systems: Exploring the
known, the unknown and the unknowable American Mathematical Society,
vol. 40, pp. 3-19, 2003.
[37] Thomas Wagner, Alan Garvey, and Victor Lesser, "Satisficing
evaluation functions: The heart of the new design-to-criteria paradigm,"
UMASS Department of Computer Science, USA, Technical Report TR-96-82,
Nov 1996.
[38] W. Brian Arthur. Introduction: Process and Emergence in the
Economy . In: The Economy as an Evolving Complex System II, eds. W.
Brian Arthur, Steven Durlauf, and David A. Lane. Reading, Mass, USA:
Addison-Wesley Pub. Co, 1997.pp. 1-04.
[39] W. C. Stirling, M. A. Goodrich, and D. J. Packard, Satisficing
Equilibria: A Non-Classical Theory of Games and Decisions Autonomous
CIS Technical Report 070305A March 2007
Agents and Multi-Agent Systems, vol. 5, pp. 305-328, Sep, 2002.
[40] Wynn C. Stirling. Satisficing Games and Decision Making: with
applications to engineering and computer science, USA: Cambridge
University Press, 2003.
[41] Wynn C. Stirling and Michael A. Goodrich, Satisficing games
Information Sciences: an International Journal, vol. 114, pp. 255-280, Mar,
1999.
[42] Youngse Kim , Satisficing and Optimality in 2 x 2 Common
Interest Games Economic Theory, vol. 13, pp. 365-375, Feb, 1999.
[43] Yu-Chi Ho , Heuristics, rules of thumb, and the 80/20
proposition IEEE Transactions on Automatic Control, vol. 39, pp. 1025-
1027, May, 1994.
[44] Yu-Chi Ho, On the numerical solutions of stochastic optimization
problem IEEE Transactions on Automatic Control, vol. 42, pp. 727-729,
1997.
page 5 of 5